Higgs Realtime Duplex: a 45B MoE model that talks, watches, and works in the background
Boson's first full-duplex interactive omni model: it listens, watches, and talks at the same time in 160ms chunks, renders its own face, and calls tools or delegates to a bigger model without pausing the conversation.
Why it matters
A mid-size 45B MoE front model that keeps a live conversation going while background work happens — the first full-duplex design that treats talking and thinking as parallel streams.
What you could build with it
A startup building a voice support agent could put Higgs Realtime Duplex in front of callers so the agent keeps talking naturally while it looks up an account, runs checks, and hands hard cases to a larger model in the background.
Does it hold up?
Too early to judge: announced Nov 5 as an API service, the benchmark table is Boson's own, and real-world usage reports have not appeared yet.
Built with Higgs Realtime Duplex
- Boson AI introduces a 45B model that talks and renders a facearticle · Coverage of the Nov 5 launch and the cost target of an hour of model time at a tenth of a person's hour.
- pipecat-boson: use Higgs Realtime with Pipecatpypi · Boson-maintained Pipecat integration exposing Higgs Realtime as a speech-to-speech service, with no separate STT, LLM, or TTS needed.
- Higgs Realtime: cost-efficient, real-time speech-to-speech modelarticle · The earlier Higgs Realtime model page: OpenAI Realtime API-compatible, priced at $0.0023/min audio in.
Learn more
First spotted on article: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73NVIDIA Isaac Lab 3.0 GA + Isaac OS open runtime + Agile ONE SN humanoidMajor Isaac platform expansion at GTC Berlin Oct 21: Isaac Lab 3.0 is generally available as the open-source robot…models · JEV 0.73NVIDIA Nemotron 3.5 Retriever collection: Embed 8B/1B, Rerank 8B, Parse 631MNVIDIA's Oct 21 collection drop for agentic retrieval: Nemotron-3.5-Embed in 8B BF16, 1B BF16 and 1B NVFP4…models · JEV 0.72Kimi K3.1 Pro: Moonshot's 2.8T open-weight flagship officially launchesMoonshot AI officially launched Kimi K3.1 Pro on Oct 23, a mixture-of-experts model with 2.8 trillion total / 104…models · JEV 0.72
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.